Comparison of Two Groups for Binary Outcomes and Test for Difference or Superiority

Abstract

In medical research, the result of an intervention is frequently measured by a binary endpoint as, for example therapy success, therapy response, or improvement of symptoms by a pre-defined extent. When comparing two treatments in clinical trials with binary outcome, this is done based on the related rates. Various measures for expressing treatment group differences are used the most popular ones being the risk difference, the risk ratio, and the odds ratio. For these measures and two-group comparisons with assessment of difference or superiority, sample size calculation methods are given for asymptotic tests (chi-square test, logistic regression) as well as for the exact unconditional Fisher-Boschloo test.